Source Code for Module ivs.aux.xmlparser

 1  # -*- coding: utf-8 -*- 
 2  """ 
 3  Convert a XML page to a dictionary 
 4  """ 
 5   
 6  import xml.parsers.expat 
 7  import urllib 
 8   
 9 -class XMLParser: 
10      """ 
11      Convert a XML page to a dictionary 
12      """ 
13 -    def __init__(self,text,logger=None): 
14          self.text = text 
15          self.content = {} 
16          self.elem = [] 
17          self.logger = logger 
18          p = xml.parsers.expat.ParserCreate() 
19          p.StartElementHandler = self.start_element 
20          p.EndElementHandler = self.end_element 
21          p.CharacterDataHandler = self.char_data 
22          if logger: 
23              logger.debug("XMLParser initialised") 
24          p.Parse(text) 
25           
26 -    def start_element(self,name,attrs): 
27          """ 
28          If a new element is found, at it to the dictionary 
29          """ 
30          if len(self.elem)==0: 
31              self.content[name] = {} 
32              self.elem.append(name) 
33          else: 
34              curdict = self.content 
35              for ie in self.elem: 
36                  curdict = curdict[ie] 
37              if not name in curdict.keys(): 
38                  curdict[name] = {} 
39              if not self.elem[-1]==name: 
40                  self.elem.append(name) 
41              if attrs: 
42                  key = attrs[attrs.keys()[0]] 
43                  #print "START2",self.elem 
44                  #print "START3",name,attrs,key,curdict[name].keys() 
45                  if key not in curdict[name].keys(): 
46                      curdict[name][key] = {} 
47                      self.elem.append(key) 
48       
49 -    def end_element(self,name): 
50          """ 
51          Remove the element from the queu (and everything after it) 
52           
53          @parameter name: designation of the element 
54          @type name: string 
55          """ 
56          #-- search for last occurence of 'name', and delete everything after it 
57          if len(self.elem)>1: 
58              index = len(self.elem) - 1 - self.elem[::-1].index(name) 
59              self.elem = self.elem[:index] 
60       
61 -    def char_data(self,data): 
62          """ 
63          Add the value of an element to the dictionary with its designation. 
64           
65          @parameter data: value of the element 
66          @type data: string 
67          """ 
68          #-- go to the final dictionary (most nested) 
69          curdict = self.content 
70          for ie in self.elem[:-1]: 
71              curdict = curdict[ie] 
72          #-- try to make it a float, if possible 
73          try: 
74              data = float(data) 
75          except: 
76              pass 
77          #-- if there is already a value for this element, make it a list and 
78          #   and the value, or just add the value. 
79          if curdict[self.elem[-1]] and not isinstance(curdict[self.elem[-1]],list): 
80              curdict[self.elem[-1]] = [curdict[self.elem[-1]]] 
81              curdict[self.elem[-1]].append(data) 
82          elif curdict[self.elem[-1]]: 
83              curdict[self.elem[-1]].append(data) 
84          else: 
85              curdict[self.elem[-1]] = data 
86          if self.logger: 
87              self.logger.debug("... %s: %s"%(self.elem,data)) 
88